An image lens distortion compensating technology is a technology that compensates a distorted image of a lens having a large viewing angle and is widely used in an endoscope or automobile field.
In particular, in the case of the automobile field, a method for verifying the vicinity of a vehicle includes a method for verifying the vicinity of the vehicle with naked eyes and a method for verifying the vicinity of the vehicle through a side mirror. However, according to a vehicle type, there is a vehicle in which the vicinity of the vehicle may be easily verified with the naked eyes or through the side mirror or a vehicle in which the vicinity of the vehicle may not be easily verified. In particular, in the case of a large-sized vehicle, an area which can not be confirmed only by the naked eye or the side mirror is relatively larger than that of a small-sized vehicle.
Therefore, in recent years, a technology has been developed, which uses a rear-view camera mounted on a vehicle or photographs a surrounding environment through a camera installed in each of front, rear, left, and right directions of the vehicle and outputs a peripheral image of the vehicle by combining photographed images. As such, the cameras installed in the front, rear, left, and right directions of the vehicle primarily adopt a wide-angle lens having the large viewing angle and a fish-eye lens which is a super wide angle lens having the viewing angle more than 180° is also used. However, in the case of an image photographed by using the lens having the large viewing angle like the wide-angle lens and the fish-eye lens, a refractive index increases toward a peripheral area of the image instead of providing a wide viewing angle, thereby showing an image having a crude distortion. Therefore, a scheme is required, which compensates the distortion of the image photographed by using the lens having the large viewing angle like the wide-angle lens and the fish-eye lens.
FIG. 1 is a diagram illustrating a process in which an image distortion compensating device in the related art compensates an image by using a mathematical model.
Referring to FIG. 1, the image distortion compensating device in the related art obtains a compensated pixel coordinate by substituting each pixel coordinate of a distorted image 110 in a predetermined mathematical expression and thereafter, arranges the obtained pixel coordinates and acquires a compensated image 120.
Such a method has a problem that since calculation needs to be performed with respect to all pixels, implementation complexity is high and a memory having a large capacity is required. Further, as the image is compensated by applying a fixation distortion compensation rate, there are a problem that a pedestrian 111 which exists in the distorted image 110 disappears in the compensated image 120 and a blind spot is thus generated and a problem that a perspective error in which the compensated image seems to be closer than an actual distance occurs.
FIG. 2 is a diagram illustrating a process in which the image distortion compensating device in the related art compensates the image by using a lookup table.
Referring to FIG. 2, the image distortion compensating device in the related art first obtains a changed pixel coordinate of an input image by applying a calibration algorithm to the input image to generate an input table 210. Thereafter, when all pixels are reconfigured, a result coordinate value to be moved is acquired. Thereafter, a revision table 230 is generated and stored through the result coordinate value. Thereafter, an output table 220 is generated, which is reconfigured by making the input table 210 correspond to the revision table 230. Thereafter, image distortion is compensated through the output table 220.
Such a method also has a problem that the memory having the large capacity is required because all pixel values after and before revision need to be stored. Further, there is also a problem in that since the image is enlarged to be larger than the existing image size in order to compensate the image with a distorted degree, the blind spot is generated and the perspective error in which the compensated image seems to be closer than the actual distance occurs.